COLM poster: training models to switch ambiguity strategies per context
JonathanBerant · x · 2026-10-07
Jonathan Berant will present his paper (arXived in December) at COLM's 11am poster session. Core idea: prompts can be ambiguous, and how to handle ambiguity is context/user dependent — sometimes the right move is asking a clarifying question, sometimes giving multiple answers, and sometimes just guessing. The work explores whether models can be trained to adapt their strategy to the context.
More from Research
- 3DGS tracking upgraded: 120fps global-shutter camera doubles pose updates — Scobleizer · 2026-10-07
- New study argues life on Earth didn't have one origin, but two — skdh · 2026-10-07
- COLM paper: a few misaligned LLM agents can sway aligned majorities — AccBalanced · 2026-10-07
- Sharing one KV cache across recursions improves looped Transformers, preprint finds — yoavartzi · 2026-10-07
- Fourier transforms pushed to O(N log(N)^0.9999...), nearly closing the log gap — burny_tech · 2026-10-07
- Stanford OVAL brings two papers to COLM2026: SatIR and DataSTORM — stanfordnlp · 2026-10-07